Charlottesville Region HMDA
Lee LeBoeuf
10/15/2021
Data Source
- These data come from the Consumer Financial Protection Bureau and Federal Financial Institutions Examinations Council. All data were collected as part of the Home Mortgage Disclosure Act (HMDA).
- New data is released biannually on a year and half delay from when it was collected. New data can be downloaded from the Federal Financial Institutions Examinations Council here.
- Data from 2017 and earlier can be downloaded from the Consumer Financial Protection Bureau here
- Data presented here are from 2007-2020 (mostly tract averages across those years).
Data specifics
- The HMDA requires financial institutions to maintain, report, and publicly disclose information about mortgages.
- New data is released annually, typically in the summer months, on a year delay (2020 data was released in June of 2021).
- Areas included in these data: Accomack (51001) and Northampton (51131)
- The data show below include only home purchase loans, meaning that mortgages for home improvement or refinancing are excluded.
- An important limitation to note is that the FFEIC allows banks to submit data with an error threshold as high as 10% for community banks and credit unions. The loan amounts and applicant income values appear prone to entry error at the individual level--a high proportion of values seemed too high to be realistic. Therefore, the tract level summaries of these variables (median loan amounts and median applicant income in the tract) should be interpreted with extreme caution.
Variable descriptions
meta %>%
filter(su_tract == 1) %>%
select(varname, about) %>% as.list()## $varname
## [1] "census_tract" "year"
## [3] "conventional" "fha_insured"
## [5] "va_guaranteed" "usda_guaranteed"
## [7] "home_purchase" "home_improve"
## [9] "refinance" "cash_out_refi"
## [11] "other_purpose" "purposeNA"
## [13] "req_preapproval" "noreq_preapproval"
## [15] "originated_loan" "approvedApp_notAccepted"
## [17] "app_denied" "app_withdrawn"
## [19] "fileclosed_incomplete" "purchased_loan"
## [21] "denied_preapproval" "approve_preapproval"
## [23] "appRace_AIAN" "appRace_Asian"
## [25] "appRace_Black" "appRace_HawPI"
## [27] "appRace_White" "appRace_missing"
## [29] "appRace_NA" "appRace_multiracial"
## [31] "appEth_HisLat" "appMale"
## [33] "appFemale" "appsex_missing"
## [35] "highCost_mortgages" "nonHighCost_mortgages"
## [37] "firstlien_secured" "sublien_secured"
## [39] "avg_loan_amount" "med_loan_amount"
## [41] "avg_app_income" "median_app_income"
## [43] "median_income_accepted_app" "white_denial_rate"
## [45] "black_denial_rate" "hislat_denial_rate"
## [47] "loans_per_units" "perc_conventional"
## [49] "perc_govern_backed" "sum_mortgage_dollars_in000s"
## [51] "avg_homepurchase_loanamount" "med_homepurchase_loanamount"
## [53] "total_apps" "perc_app_missingRace"
## [55] "overall_denial_rate" "perc_white_apps"
## [57] "perc_black_apps" "perc_hislat_apps"
## [59] "tract_population" "minority_population"
## [61] "median_family_income" "tract_to_msamd_income"
## [63] "tract_owner_occupied_units" "tract_one_to_four_family_homes"
##
## $about
## [1] "11-digit tract code"
## [2] "The year"
## [3] "The number of conventional loans"
## [4] "The number of loans insured by the Federal Housing Administration"
## [5] "The number of VA guaranteed loans"
## [6] "The number of USDA guaranteed loans"
## [7] "The number of applications or covered loans for home purchase"
## [8] "The number of applications or covered loans for home improvement"
## [9] "The number of applications or covered loans for refinancing"
## [10] "The number of applications or covered loans for cash-out refinancing"
## [11] "The number of applications or covered loans for some other purpose"
## [12] "The number of applications or covered loans for a not applicable purpose"
## [13] "The number of applications or covered loans that requested preapproval of a home purchase loan under a preapproval program"
## [14] "The number of applications or covered loans that did not request preapproval of a home purchase loan under a preapproval program"
## [15] "The number of originated loans"
## [16] "The number of approved but not accepted applications"
## [17] "The number of denied applications"
## [18] "The number of applications withdrawn by the applicant"
## [19] "The number of application files closed for incompleteness"
## [20] "The number of purchased loans"
## [21] "The number of applications for which preapproval was denied"
## [22] "The number of applications for which preapproval request was approved but not accepted"
## [23] "The number of applicants who self-identified as American Indian or Alaska Native"
## [24] "The number of applicants self-identified as Asian including Asian Indian, Chinese, Filipino, Japanese, Korean, Vietnamese, or other Asian countries"
## [25] "The number of applicants who self-identified as Black"
## [26] "The number of applicants who self-identified as Native Hawaiian or Other Pacific Islander"
## [27] "The number of applicants who self-identified as White"
## [28] "The number of applicants who did not provide racial demographic information in their application by mal, internet, or telephone"
## [29] "The number of applicants for which racial identity is not applicable"
## [30] "The number of applicants who selected multiple racial identities"
## [31] "The number of applicants/borrower is Hispanic or Latino"
## [32] "The number of applicants who are male"
## [33] "The number of applicants who are female"
## [34] "The number of applicants who did not provide their sex on their application"
## [35] "The number of covered loans that are high-cost mortgages"
## [36] "The number covered loans that are not a high cost mortgages"
## [37] "The number of covered loans/applications secured by the first lien"
## [38] "The number of covered loans/applications secured by a subordinate lien"
## [39] "The tract average loan amount. This variable seems prone to a high amount of entry error in the individual-level data, so users should be cautious about using the tract-level summary."
## [40] "The tract median loan amount. This variable seems prone to a high amount of entry error in the individual-level data, so users should be cautious about using the tract-level summary."
## [41] "The tract average applicant income. This variable seems prone to a high amount of entry error in the individual-level data, so users should be cautious about using the tract-level summary."
## [42] "The tract median applicant income. This variable seems prone to a high amount of entry error in the individual-level data, so users should be cautious about using the tract-level summary."
## [43] "The tract median income of accepted applications"
## [44] "The tract denial rate for white applicants"
## [45] "The tract denial rate for black applicants"
## [46] "The tract denial rate for Hispanic or Latino applicants"
## [47] "The number of loans per the number of one to four family homes in the tract"
## [48] "The percent of approved applications that were conventional"
## [49] "The percent of approved applications that were backed by the USDA, VA, or FHA"
## [50] "The total amount of the covered loans, or the amounts applied for in the tract"
## [51] "The tract average loan amount for originated loans"
## [52] "The tract median loan amount"
## [53] "The total number of applications in the tract"
## [54] "The percent of applications in which applicants did not disclose their racial identity"
## [55] "The tract denial rate for all applicants"
## [56] "The percent of applications in the tract from White applicants"
## [57] "The percent of applications from Black applicants"
## [58] "The percent of applications from Hispanic or Latino applicants"
## [59] "The population in the tract"
## [60] "The percentage of minority population to total ppulation for tract, rounded to two decimals places"
## [61] "The median family income in the tract"
## [62] "The percentage of tract median family income compared to MSA/MD median family income"
## [63] "The number of dwellings, including individual condominiums, that are lived in by the owner"
## [64] "The number of dwellings that are built to houses with fewer with than 5 families"glimpse(eastdat)## Rows: 160
## Columns: 53
## $ census_tract <dbl> 51001090100, 51001090100, 51001090100, …
## $ year <int> 2012, 2013, 2014, 2015, 2016, 2017, 201…
## $ conventional <int> 64, 111, 108, 90, 120, 155, 139, 166, 2…
## $ fha_insured <int> 1, 3, 6, 7, 6, 6, 19, 14, 20, 11, 20, 1…
## $ va_guaranteed <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ usda_guaranteed <int> 0, 0, 0, 2, 1, 0, 0, 1, 6, 4, 9, 13, 8,…
## $ req_preapproval <int> 4, 3, 8, 5, 4, 8, 3, 3, 3, 0, 6, 8, 1, …
## $ noreq_preapproval <int> 30, 43, 30, 11, 30, 19, 164, 191, 255, …
## $ originated_loans <int> 47, 65, 68, 66, 89, 109, 103, 113, 167,…
## $ approvedApp_notAccepted <int> 3, 4, 4, 1, 6, 3, 1, 3, 4, 1, 2, 5, 3, …
## $ app_denied <int> 4, 25, 21, 15, 8, 21, 15, 11, 16, 14, 1…
## $ app_withdrawn <int> 4, 3, 11, 8, 16, 11, 18, 28, 22, 1, 5, …
## $ fileclosed_incomplete <int> 0, 1, 1, 3, 3, 6, 6, 4, 5, 1, 1, 4, 8, …
## $ purchased_loan <int> 10, 21, 12, 18, 18, 19, 24, 35, 44, 16,…
## $ denied_preapproval <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ approve_preapproval <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ appRace_AIAN <int> 0, 0, 0, 3, 0, 0, 0, 0, 1, 0, 0, 0, 2, …
## $ appRace_Asian <int> 0, 0, 1, 1, 1, 0, 0, 0, 4, 0, 1, 0, 1, …
## $ appRace_Black <int> 0, 0, 0, 1, 2, 2, 0, 1, 2, 5, 12, 10, 1…
## $ appRace_HawPI <int> 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, …
## $ appRace_White <int> 53, 93, 101, 88, 110, 137, 146, 153, 19…
## $ appRace_multiracial <int> 0, 0, 0, 0, 0, 1, 0, 0, 3, 0, 0, 0, 4, …
## $ appRace_missing <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ appRace_NA <int> 7, 10, 6, 6, 11, 11, 12, 22, 35, 7, 11,…
## $ appEth_HispLat <int> 0, 1, 4, 1, 0, 4, 2, NA, NA, 0, 0, 1, 1…
## $ appMale <int> 37, 63, 65, 62, 86, 95, 95, 96, 128, 39…
## $ appFemale <int> 20, 33, 41, 36, 28, 50, 50, 61, 80, 20,…
## $ appsex_missing <int> 4, 13, 5, 7, 15, 13, 10, 15, 15, 4, 2, …
## $ highCost_mortgages <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ nonHighCost_mortgages <int> 68, 119, 117, 111, 140, 169, 60, 69, 91…
## $ firstlien_secured <int> 58, 97, 104, 93, 121, 149, 165, 192, 25…
## $ sublien_secured <int> 0, 1, 1, 0, 1, 1, 2, 2, 2, 0, 0, 1, 2, …
## $ median_income_000s <dbl> 130.0, 106.5, 107.0, 110.5, 118.5, 117.…
## $ med_loan_amount_000s <dbl> 164.0, 162.0, 143.0, 170.0, 177.0, 155.…
## $ total_apps <int> 68, 119, 117, 111, 140, 169, 167, 194, …
## $ overall_denial_rate <dbl> 0.05882353, 0.21008403, 0.17948718, 0.1…
## $ perc_white_apps <dbl> 77.94118, 78.15126, 86.32479, 79.27928,…
## $ perc_black_apps <dbl> 0.0000000, 0.0000000, 0.0000000, 0.9009…
## $ perc_hislat_apps <dbl> 0.0000000, 0.8403361, 3.4188034, 0.9009…
## $ white_denial_rate <dbl> 0.05660377, 0.21505376, 0.17821782, 0.1…
## $ black_denial_rate <dbl> NA, NA, NA, 1.0000000, 0.0000000, 0.500…
## $ hislat_denial_rate <dbl> NA, 0.00, 0.25, 0.00, NA, 0.25, 0.00, N…
## $ median_income_accepted_app <dbl> 134.0, 107.0, 114.0, 118.0, 119.0, 119.…
## $ loans_per_units <dbl> 0.011573504, 0.016005910, 0.016744644, …
## $ perc_conventional <dbl> 97.87234, 93.84615, 94.11765, 86.36364,…
## $ perc_govern_backed <dbl> 2.127660, 6.153846, 5.882353, 13.636364…
## $ sum_mortgage_dollars_in000s <int> 9005, 12187, 12468, 11616, 18309, 20637…
## $ perc_app_missingRace <dbl> 11.764706, 13.445378, 7.692308, 10.8108…
## $ tract_population <int> 2941, 2941, 2941, 2941, 2941, 2930, 293…
## $ minority_population <dbl> 6.02, 6.02, 6.02, 6.02, 6.02, 6.62, 6.6…
## $ median_family_income <int> 52600, 51600, 52000, 52700, 52300, 5330…
## $ tract_owner_occupied_units <int> 1323, 1323, 1323, 1323, 1323, 1095, 109…
## $ tract_one_to_four_family_homes <int> 4061, 4061, 4061, 4061, 4061, 4158, 415…eastdat[which(eastdat$year == 2020),] %>% select(total_apps, overall_denial_rate, white_denial_rate, black_denial_rate, hislat_denial_rate, perc_conventional, perc_govern_backed) %>%
select(where(~is.numeric(.x))) %>%
as.data.frame() %>%
stargazer(., type = "text", title = "Summary Statistics", digits = 2,
summary.stat = c("mean", "sd", "min", "median", "max"))##
## Summary Statistics
## ======================================================
## Statistic Mean St. Dev. Min Median Max
## ------------------------------------------------------
## total_apps 121.27 78.66 26 105 258
## overall_denial_rate 0.11 0.04 0.06 0.11 0.19
## white_denial_rate 0.09 0.04 0.04 0.07 0.18
## black_denial_rate 0.27 0.26 0 0.3 1
## hislat_denial_rate 0.06 0.10 0.00 0.00 0.17
## perc_conventional 70.86 11.58 53.85 68.00 88.62
## perc_govern_backed 29.14 11.58 11.38 32.00 46.15
## ------------------------------------------------------Visual distribution
- The following distributions include data only from the year 2020.
longdat <- eastdat[which(eastdat$year == 2020),] %>% select(c(census_tract, total_apps, overall_denial_rate, white_denial_rate, black_denial_rate, hislat_denial_rate, perc_conventional, perc_govern_backed)) %>% pivot_longer(-census_tract, names_to = "measure", values_to = "value")
longdat$measure <- factor(longdat$measure,
levels = c("total_apps", "overall_denial_rate", "white_denial_rate", "black_denial_rate", "hislat_denial_rate", "perc_conventional", "perc_govern_backed"))
longdat %>%
ggplot(aes(x = value, fill = measure)) +
scale_fill_viridis(option = "plasma", discrete = TRUE, guide = FALSE) +
geom_histogram() +
facet_wrap(~measure, scales = "free")## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 8 rows containing non-finite values (stat_bin).
meta %>%
filter(varname %in% c("total_apps", "overall_denial_rate", "white_denial_rate", "black_denial_rate", "hislat_denial_rate", "perc_conventional", "perc_govern_backed")) %>%
mutate(label = paste0(varname, ": ", about)) %>%
select(label) %>%
as.list()$label [1] "white_denial_rate: The tract denial rate for white applicants"
[2] "black_denial_rate: The tract denial rate for black applicants"
[3] "hislat_denial_rate: The tract denial rate for Hispanic or Latino applicants"
[4] "perc_conventional: The percent of approved applications that were conventional"
[5] "perc_govern_backed: The percent of approved applications that were backed by the USDA, VA, or FHA" [6] "total_apps: The total number of applications in the tract"
[7] "overall_denial_rate: The tract denial rate for all applicants"
Mapping the data
Total number of applications in 2020
mapdat2020 <- mapdat[which(mapdat$year == 2020),]
pal <- colorNumeric("plasma", reverse = TRUE, domain = mapdat2020$total_apps)
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat2020,
fillColor = ~pal(mapdat2020$total_apps),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat2020$census_tract, "<br>",
"Number of applications: ", mapdat2020$total_apps)
) %>%
addLegend("bottomright", pal = pal, values = (mapdat2020$total_apps),
title = "Total number of <br>mortgage applications <br> in 2020", opacity = 0.7)Average mortage application denial rates from 2007-2020
- Each of the following maps show the average denial rates for the years 2007-2020
Overall denial rate
mapdat <- mapdat %>%
group_by(census_tract) %>%
mutate(avg_overall_denial_rate = mean(na.omit(overall_denial_rate))) %>%
ungroup()
pal <- colorNumeric("plasma", reverse = TRUE, domain = mapdat$avg_overall_denial_rate)
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat,
fillColor = ~pal(mapdat$avg_overall_denial_rate),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat$census_tract, "<br>",
"Average overall app denial rate from 2007-2020: ", round(mapdat$avg_overall_denial_rate, 2))
) %>%
addLegend("bottomright", pal = pal, values = (mapdat$avg_overall_denial_rate),
title = "Average overall <br>app denial rate <br>from 2007-2020", opacity = 0.7)White denial rate
mapdat <- mapdat %>%
group_by(census_tract) %>%
mutate(avg_white_denial_rate = mean(na.omit(white_denial_rate))) %>%
ungroup()
pal <- colorNumeric("plasma", reverse = TRUE, domain = mapdat$avg_white_denial_rate)
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat,
fillColor = ~pal(mapdat$avg_white_denial_rate),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat$census_tract, "<br>",
"Average White app denial rate from 2007-2020: ", round(mapdat$avg_white_denial_rate, 2))
) %>%
addLegend("bottomright", pal = pal, values = (mapdat$avg_white_denial_rate),
title = "Average White <br>app denial rate <br>from 2007-2020", opacity = 0.7)Black denial rate
mapdat <- mapdat %>%
group_by(census_tract) %>%
mutate(avg_black_denial_rate = mean(na.omit(black_denial_rate))) %>%
ungroup()
pal <- colorNumeric("plasma", reverse = TRUE, domain = (mapdat$avg_black_denial_rate))
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat,
fillColor = ~pal((mapdat$avg_black_denial_rate)),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat$census_tract, "<br>",
"Average Black app denial rate from 2007-2020: ", round(mapdat$avg_black_denial_rate, 2))
) %>%
addLegend("bottomright", pal = pal, values = (mapdat$avg_black_denial_rate),
title = "Average Black <br>app denial rate <br>from 2007-2020", opacity = 0.7)Hispanic/Latino denial rate
mapdat <- mapdat %>%
group_by(census_tract) %>%
mutate(avg_hislat_denial_rate = mean(na.omit(hislat_denial_rate))) %>%
ungroup()
pal <- colorNumeric("plasma", reverse = TRUE, domain = (mapdat$avg_hislat_denial_rate))
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat,
fillColor = ~pal((mapdat$avg_hislat_denial_rate)),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat$census_tract, "<br>",
"Average His/Lat app denial rate from 2007-2020: ", round(mapdat$avg_hislat_denial_rate, 2))
) %>%
addLegend("bottomright", pal = pal, values = (mapdat$avg_hislat_denial_rate),
title = "Average Hispanic/Latino <br>app denial rate <br>from 2007-2020", opacity = 0.7)Government backed mortgages
- Government back mortages are mortgages insured by the USDA, VA, or FHA
Percent of government backed loans over time
shape <- readRDS('eastshore_tracts.RDS')
shape <- shape %>% dplyr::rename(census_tract = GEOID)
animatemapdat <- merge(shape, eastdat, by = 'census_tract', all.x = T)
animatemapdat <- st_as_sf(animatemapdat)
animatemapdat$year = as.numeric(animatemapdat$year)
animatemapdat <- animatemapdat %>% filter_at(vars(NAME, geometry, perc_govern_backed),all_vars(!is.na(.)))
cville1 <-
ggplot(animatemapdat) +
geom_sf(aes(fill = perc_govern_backed), color = "black", alpha = .9, na.rm = TRUE) +
scale_fill_fermenter(palette = "Blues", direction = 1, type = "seq", n.breaks = 7) +
theme_void() +
guides(fill = guide_colourbar(title.position="top", title.hjust = 0.5, barwidth = 1)) +
labs(fill = "Percent of mortgages backed by gov.", title = 'Year:{frame_time}',
caption = "Percent of approved mortgage apps backed by the USDA, VA, or FHA") +
transition_time(as.integer(year)) +
ease_aes('linear')
animate(cville1, fps = 1, nframes = 13)
Average percent of government backed loans from 2007-2020
mapdat <- mapdat %>%
group_by(census_tract) %>%
mutate(avg_percgovbacked = mean(na.omit(perc_govern_backed))) %>%
ungroup()
pal <- colorNumeric("plasma", reverse = TRUE, domain = mapdat$avg_percgovbacked)
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat,
fillColor = ~pal(mapdat$avg_percgovbacked),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat$census_tract, "<br>",
"Average % of gov-backed mortgages <br> from 2006-2020: ", round(mapdat$avg_percgovbacked, 2))
) %>%
addLegend("bottomright", pal = pal, values = (mapdat$avg_percgovbacked),
title = "Average % of <br>gov-backed mortgages <br>from 2007-2020", opacity = 0.7)Percent of government backed loans in 2020
mapdat2020 <- mapdat[which(mapdat$year == 2020),]
pal <- colorNumeric("plasma", reverse = TRUE, domain = mapdat2020$perc_govern_backed)
leaflet() %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons(data = mapdat2020,
fillColor = ~pal(mapdat2020$perc_govern_backed),
weight = 1,
opacity = 1,
color = "white",
fillOpacity = 0.6,
highlight = highlightOptions(
weight = 2,
fillOpacity = 0.8,
bringToFront = T
),
popup = paste0("GEOID: ", mapdat2020$census_tract, "<br>",
"Percent of gov-backed <br>mortages in 2020: ", round(mapdat2020$perc_govern_backed, 2))
) %>%
addLegend("bottomright", pal = pal, values = (mapdat2020$perc_govern_backed),
title = "Percent of <br>gov-backed <br>mortages in 2020", opacity = 0.7)